Bug 21772 - john is unusable in i586 because he asks for SSE2 even if the CPU has SSE2
Summary: john is unusable in i586 because he asks for SSE2 even if the CPU has SSE2
Status: RESOLVED FIXED
Alias: None
Product: Mageia
Classification: Unclassified
Component: RPM Packages (show other bugs)
Version: 6
Hardware: i586 Linux
Priority: Normal critical
Target Milestone: ---
Assignee: QA Team
QA Contact:
URL:
Whiteboard: MGA6-32-OK MGA6-64-OK
Keywords: advisory, validated_update
Depends on:
Blocks:
 
Reported: 2017-09-26 15:32 CEST by José Jorge
Modified: 2017-11-19 11:24 CET (History)
4 users (show)

See Also:
Source RPM: john
CVE:
Status comment:


Attachments

Description José Jorge 2017-09-26 15:32:43 CEST
1. Install john
2. Run john

It says :

Sorry, SSE2 is required for this build

/proc/cpuinfo in my system says I have it :

cat /proc/cpuinfo | grep sse
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov clflush dts acpi mmx fxsr sse sse2 ss tm pbe nx bts eagerfpu est tm2
Rémi Verschelde 2017-09-26 15:36:15 CEST

Assignee: bugsquad => matteo.pasotti

Comment 1 José Jorge 2017-09-28 15:18:32 CEST
I have rebuilt the package in this system, and it activates SSE2 support and works. So I am testing a rebuild of it in our bs to ensure it is not just a build problem.

CC: (none) => lists.jjorge

Comment 2 José Jorge 2017-09-28 15:43:54 CEST
This test build shows the AVX instructions are detected and activated in our build system. So the SSE2 message is probably about AVX instructions missing.

I have not found a parameter to disable SSE, SSE2 and AVX in i586 builds of john.
The workaround for now is to rebuild the package in the target hardware.

Status: NEW => ASSIGNED

Comment 3 José Jorge 2017-11-06 16:38:02 CET
Thanks to Peter Semiletov, I've found a configure parameter to disable CPU detection. Pushed version 1.8.0-5.2 to testing.

Assignee: matteo.pasotti => qa-bugs

Comment 4 Lewis Smith 2017-11-06 21:57:38 CET
Not enough info for the advisory. The bug RPMs link shows nothing. Is/are the package(s) in updates_testing ?
This is a rare bug that looks 32-bit *only*; no 64-bit test required. Please say if this is assumption incorrect.
Comment 5 José Jorge 2017-11-07 10:07:35 CET
(In reply to Lewis Smith from comment #4)
> Not enough info for the advisory. The bug RPMs link shows nothing. Is/are
> the package(s) in updates_testing ?

Sorry, I will add the suggested advisory.

> This is a rare bug that looks 32-bit *only*; no 64-bit test required. Please
> say if this is assumption incorrect.

Incorrect : you can try john on an old x86_64 system without SSE3 or AVX instructions. It will fail. So it must be tested in both arches.
I have tested in my i586 system (Pentium M 740 - only SSE2) and the update fixed the bug.
Comment 6 José Jorge 2017-11-07 10:12:15 CET
Suggested advisory:
========================

Updated john packages fix crash at startup with old CPUs:

john was built with automatic cpu detection. So it enabled SSE3 and AVX instructions even for 32 bit packages.
This update limits the autorised cpu instructions for both i586 and x86_64.

========================

Updated packages in core/updates_testing:
========================
john-1.8.0-5.2.mga6

Source RPMs: 
john-1.8.0-5.2.mga6
José Jorge 2017-11-08 14:22:52 CET

Whiteboard: (none) => MGA6-32-OK

Comment 7 Len Lawrence 2017-11-09 20:38:59 CET
Don't know if this hardware is any use for testing.  10 year old Dell XPS laptop.
CPU:       Dual core Intel Core2 Duo T9300 (-MCP-)
Just resurrected it and installed Mageia 6.

$ cat /proc/cpuinfo
...................
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx lm constant_tsc arch_perfmon pebs bts rep_good nopl aperfmperf eagerfpu pni dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm sse4_1 lahf_lm tpr_shadow vnmi flexpriority dtherm ida

Pre update john launches OK from the commandline.

Likewise after the update.
So, not much help.

CC: (none) => tarazed25

Comment 8 José Jorge 2017-11-12 22:08:42 CET
I have tested in a 8 years old AMD Athlon 5600+. It has no SSE3. Still the bug is not triggered before the update, so it seems it was really only in i586.

The update does not hurt, so validating in 64 bit.

Whiteboard: MGA6-32-OK => MGA6-32-OK MGA6-64-OK

Lewis Smith 2017-11-18 22:29:27 CET

CC: (none) => lewyssmith, sysadmin-bugs
Keywords: (none) => advisory, validated_update

Comment 9 Mageia Robot 2017-11-19 11:24:22 CET
An update for this issue has been pushed to the Mageia Updates repository.

https://advisories.mageia.org/MGAA-2017-0111.html

Resolution: (none) => FIXED
Status: ASSIGNED => RESOLVED


Note You need to log in before you can comment on or make changes to this bug.